Geri Dön

E-ticaret firmalarında kargo toplama operasyonlarının mekansal analizi ve optimizasyonu: Esenyurt, İstanbul örneği

Spatial analysis and optimization of e-commerce company's parcel collection operations: The case of Esenyurt, İstanbul

PDF İndir

  1. Tez No: 958931
  2. Yazar: FATMA REYYAN SARIKAYA
  3. Danışmanlar: DOÇ. DR. ADALET DERVİŞOĞLU
  4. Tez Türü: Yüksek Lisans
  5. Konular: Jeodezi ve Fotogrametri, Geodesy and Photogrammetry
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2025
  8. Dil: Türkçe
  9. Üniversite: İstanbul Teknik Üniversitesi
  10. Enstitü: Lisansüstü Eğitim Enstitüsü
  11. Ana Bilim Dalı: Geomatik Mühendisliği Ana Bilim Dalı
  12. Bilim Dalı: Geomatik Mühendisliği Bilim Dalı
  13. Sayfa Sayısı: 85

Özet

Bu çalışma, İstanbul'un nüfus ve sipariş yoğunluğu bakımından öne çıkan Esenyurt ilçesinde yürütülen kargo toplama operasyonlarının mekânsal ve operasyonel açıdan değerlendirilmesini amaçlamaktadır. Günümüzde e-ticaretin hızlı gelişimiyle birlikte, kentsel lojistik süreçlerin daha verimli, esnek ve veri temelli yapılara dönüştürülmesi gerekliliği artmıştır. Bu doğrultuda, çalışmada Coğrafi Bilgi Sistemleri (CBS) kullanılarak siparişlerin konumsal dağılımı analiz edilmiş, Araç Rotalama Problemi (VRP) çözüm modelleriyle toplama rotaları yapılandırılmış ve operasyonel ölçütler üzerinden senaryo bazlı değerlendirmeler yapılmıştır. Analiz sürecinin ilk aşamasında, sahadan elde edilen sipariş verileri üzerinde Kernel Yoğunluk Tahmini (KDE) yöntemi kullanılarak desi bazlı sipariş kümelenmeleri ortaya konulmuş ve bu kümeler operasyonel planlamaya temel olacak şekilde görselleştirilmiştir. İkinci aşamada, ArcGIS Pro yazılımının Network Analyst modülü kullanılarak çeşitli operasyonel kısıtlar altında rota planlamaları yapılmıştır. Bu kısıtlar arasında araç kapasitesi, hizmet süresi, mola aralıkları ve çalışma zamanları yer almıştır. Çalışmanın devamında, açık kaynaklı bir çözümleyici olan Google OR-Tools kütüphanesiyle, aynı veri seti üzerinde alternatif bir rota optimizasyon modeli geliştirilmiştir. Modelin girdileri, Google Maps Distance Matrix API kullanılarak oluşturulan mesafe matrisi üzerinden hazırlanmış; rota süreleri, hizmet zamanları ve yük dengesi gibi parametreler sistematik biçimde tanımlanmıştır. Python dili kullanılarak geliştirilen bu yapı, kod temelli senaryo üretimi ve esneklik açısından uygulama avantajları sunmuştur. Çalışma boyunca elde edilen bulgular, operasyonel yük dağılımı, araç kullanım verimliliği ve sistem esnekliği gibi faktörler üzerinden analiz edilmiştir. Yapılan karşılaştırmalar, farklı yazılım altyapılarının güçlü ve sınırlı yönlerini ortaya koymuş; kentsel kargo toplama süreçlerinin iyileştirilmesinde veri tabanlı yaklaşımların sunduğu katkıların daha iyi anlaşılmasını sağlamıştır. Bu çerçevede çalışma, mekânsal analizler ve algoritmik modelleme tekniklerinin, kent içi lojistik planlamaya entegre edilebileceği bir karar destek sistemi mantığına dayalı uygulama örneği sunmaktadır.

Özet (Çeviri)

The rapid expansion of the e-commerce sector has profoundly reshaped urban logistics by increasing the demand for faster, more efficient, and data-driven solutions. While the last-mile delivery segment has been widely studied and optimized in logistics literature, the earlier stage of the logistics chain — parcel collection from sellers or local warehouses to branch distribution centers — remains underexplored despite its critical impact on overall performance. This thesis investigates the parcel collection operations in Esenyurt, one of Istanbul's most densely populated and commercially active districts, through a comprehensive spatial and operational analysis. By integrating Geographic Information Systems (GIS), Vehicle Routing Problem (VRP) modeling, and Multi-Criteria Decision-Making (MCDM) techniques, the study aims to provide actionable insights into improving urban parcel logistics. Esenyurt was selected as the case study due to its strategic logistics importance. It has the highest population among Istanbul's districts and serves as a regional hub for commercial activity and warehousing, with direct access to major transportation corridors such as the E-5 and TEM highways. However, this advantage is tempered by serious logistical challenges including traffic congestion, narrow street networks, high order density, and inconsistently located collection points. These conditions make Esenyurt an ideal testbed for examining how advanced spatial and computational tools can be used to design more efficient cargo collection systems. The research was conducted using real-world data provided by a leading Turkish e- commerce firm. This data included over 100,000 parcel pickup records with associated geographic coordinates, volumetric weights (in desi), and time stamps. The study focused on November 29, 2024 — the peak operational day in terms of order volume — to simulate the system under maximum pressure and evaluate its limitations and opportunities. In the first stage of the analysis, the Kernel Density Estimation (KDE) method was employed using ArcGIS Pro to visualize and quantify the spatial clustering of parcel orders. KDE is a widely used statistical technique that transforms point-based spatial data into a continuous surface, revealing the geographic intensity of order demand. By doing so, the study identified critical hotspots where pickup activity was most concentrated. This step not only provided an intuitive visual representation of operational density but also laid the groundwork for informed route planning and resource allocation. Next, the study leveraged ArcGIS Pro's Network Analyst module to develop optimized vehicle routes based on real street networks, traffic patterns, and logistical constraints. The model incorporated a range of real-world factors, such as vehicle capacity limits, driver working hours, lunch breaks, service times per stop, and maximum route durations. A total of 17 delivery vehicles — 9 from Esenyurt and 8 from Haramidere — were modeled to handle the workload across two branches. The ArcGIS-based VRP solution attempted to assign each parcel pickup to an optimal route, ensuring minimum travel time and balanced load distribution across vehicles. The third component of the study involved an alternative modeling approach using the open-source Google OR-Tools framework, integrated within a Python environment. In this scenario, distance matrices were generated via the Google Maps Distance Matrix API, offering realistic travel time estimations between parcel locations and branch depots. The distance matrix was preprocessed and cached to improve performance, and various constraints such as vehicle capacity and service time were modeled in detail. This framework offered more flexibility and computational efficiency, allowing for rapid testing of alternative scenarios and larger datasets. Both ArcGIS and OR-Tools modeling results were evaluated based on multiple performance criteria. These included the number of successfully collected parcels, total route duration, vehicle utilization rates, and unfulfilled orders. In the ArcGIS- based scenario, approximately 66,000 out of 104,000 parcels were successfully collected, while 38,000 remained uncollected due to time or capacity constraints. This outcome highlights the need for better route design, additional resources, or alternative planning strategies. The KDE results supported these findings by pinpointing high- demand areas that could benefit from micro-depot deployment or targeted fleet increases. To compare the scenarios and select optimal solutions, Multi-Criteria Decision- Making (MCDM) techniques were applied. Instead of relying solely on a single metric such as cost or time, MCDM allowed for a more nuanced evaluation based on a combination of criteria including service coverage, efficiency, flexibility, and feasibility. This method acknowledged the complexity of urban logistics, where decision-makers must trade off between competing objectives and constraints. Through this evaluation, the study demonstrated that hybrid approaches — integrating spatial intelligence with algorithmic optimization — yield more robust and actionable solutions for parcel collection. One of the key contributions of the study is its demonstration of how spatial data science can be used to support operational planning in e-commerce logistics. The use of GIS for density mapping, network modeling, and service area analysis allowed for a clear visualization of logistical bottlenecks and opportunities. Meanwhile, computational routing with OR-Tools illustrated how scenario planning and algorithmic experimentation can aid in adapting to peak demand and systemic constraints. This research also fills a gap in both academic and applied literature by focusing on the parcel collection phase of the e-commerce logistics chain. Traditionally overshadowed by last-mile delivery, parcel collection is equally important for ensuring overall service efficiency. The findings suggest that investment in data infrastructure, spatial analysis capabilities, and optimization tools can significantly improve the performance of this often-overlooked process. Moreover, the methodology developed in this thesis is generalizable to other high-density urban areas facing similar logistical challenges. In conclusion, this thesis presents a practical and interdisciplinary framework for optimizing urban parcel collection operations. By combining real-world data, spatial analytics, route optimization, and multi-criteria evaluation, it offers both a theoretical and practical contribution to the field of urban logistics. The case study of Esenyurt serves not only as a local solution model but also as a scalable template for logistics planners working in complex urban settings. The integration of Geographic Information Systems, algorithmic routing tools, and decision-support methodologies lays the foundation for smarter, more responsive, and more sustainable logistics planning in the age of e-commerce.

Benzer Tezler

  1. Tez No

    808445

    Hibrit skuter güç grubu tasarımı

    Hybrid scooter powertrain design

    MUSTAFA KABAKÇI

    Yüksek Lisans

    Türkçe

    Türkçe

    2023

    Makine MühendisliğiYıldız Teknik Üniversitesi

    Makine Mühendisliği Ana Bilim Dalı

    DOÇ. DR. ALP TEKİN ERGENÇ

  2. Tez No

    524655

    Elektronik ticaretin hazır giyim sektöründe üreticiler ve tüketiciler açısından kullanım durumu

    Using electronic trade in ready-made sector in terms of manufacturers and consumers

    MELTEM ERDOĞAN ÇAĞLAYAN

    Yüksek Lisans

    Türkçe

    Türkçe

    2018

    Giyim EndüstrisiGazi Üniversitesi

    Giyim Endüstrisi ve Moda Tasarımı Eğitimi Ana Bilim Dalı

    YRD. DOÇ. DR. HACI FATMA ŞENER

  3. Tez No

    870515

    E-ticarete markalaşma süreci ve bir uygulama

    E-commerce branding process and a survey

    ONUR ÖZTÜRK

    Yüksek Lisans

    Türkçe

    Türkçe

    2023

    Uluslararası İlişkilerGiresun Üniversitesi

    Uluslararası Ticaret ve Lojistik Yönetimi Ana Bilim Dalı

    DOÇ. DR. SALİH MEMİŞ

  4. Tez No

    719271

    Can hyperloop technology take Turkish logistics industry to a next level?: A preliminary analysis

    Hyperloop teknolojisi Türkiye lojistik sektörünü bir sonraki seviyeye taşıyabilir mi?: Ön analiz

    GÖRKEM KEREM

    Yüksek Lisans

    İngilizce

    İngilizce

    2022

    Mühendislik Bilimleriİstanbul Teknik Üniversitesi

    İşletme Mühendisliği Ana Bilim Dalı

    PROF. DR. DİLAY ÇELEBİ

  5. Tez No

    783971

    Kargo taşımacılığı sektöründe son adım dağıtım planlama ve rota optimizasyonu

    Last mile delivery planning and route optimization in cargo transportation

    SELİN BOSTAN

    Yüksek Lisans

    Türkçe

    Türkçe

    2023

    Jeodezi ve Fotogrametriİstanbul Teknik Üniversitesi

    Bilişim Uygulamaları Ana Bilim Dalı

    DOÇ. DR. CANER GÜNEY

Geri Dön